J Popul Econ 2001) 14: 173±191 999 2001 Cyclical variations in unemployment duration Michael Rosholm Department of Economics, Aarhus School of Business Fuglesangs Alle  20, DK-8210 Aarhus V, and Centre for Labour Market and Social Research Fax: 45 86 15 51 75; e-mail: rom@cls.dk) Received: 7 December 1998/Accepted: 24 August 1999 Abstract. In this paper I study how individual unemployment durations vary over the business cycle, using unemployment spells of a sample of Danish workers. A compositional, an out¯ow, and a residual calendar-time compo- nent are identi®ed, and they all contribute to explaining the variations in un- employment duration. Based on the analysis it is concluded that long-term unemployment is a phenomenon that is associated with periods of high un- employment, but nothing should prevent the long-term unemployed ®nding jobs again as aggregate unemployment eventually starts falling. In particular, there is no evidence of negative duration dependence, not even at long dura- tions, and not when aggregate unemployment is high. JEL classi®cation: E32, J64 Key words: Unemployment duration, business cycle ¯uctuation 1. Introduction The purpose of this paper is to analyse the way individual unemployment durations vary over the business cycle. A compositional and an out¯ow e¨ect are identi®ed. The compositional e¨ect consists of cyclical variations in the observed as well as unobserved `quality' of those ¯owing into unemployment, leaving individual hazard rates unchanged. The out¯ow e¨ect a¨ects the haz- Financial support from the Danish National Research Foundation is gratefully acknowledged. I would like to thank Henning Bunzel, Anne Mùller Danù, Peter Jensen, Nick Kiefer, Lars Muus, George Neumann, Geert Ridder, Niels Westergaard-Nielsen, and two anonymous referees for helpful suggestions. Remaining errors are my own. Responsible editor: John F. Ermisch.